A comprehensive report to the National Institute of Health on the diagnosis, etiology, epidemiology, and treatment of autism indicated that early intervention has the potential of being an effective intervention (Bristol et al., 1996). In spite of this positive outlook, several research and methodological questions remain regarding time of treatment initiation, intensity of treatment and duration of treatment, random assignment, comparative treatment designs, and treatment integrity. Against this backdrop we consider the claims made by the Early Intervention Project (EIP; Lovaas, 1987, 1993; McEachin, Smith, & Lovaas, 1993). The EIP claims to produce recovery from autism in 47% of the cases and to greatly reduce its severity in an additional 42% of cases. This article evaluates the EIP against threats to internal and external validity and is found to suffer from a number of methodological problems. Based on rebuttals to criticisms of their program, the EIP authors seem unwilling to admit any methodological flaws in the sampling, design, and analysis of data of the EIP. It is recommended that parents and fair hearing officers adopt an attitude of healthy skepticism before proceeding to an unqualified endorsement of the EIP as a treatment for autism.